In 1997, a transgender woman with HIV entering a French hospital for the first time waited a median of almost four months before starting antiretroviral therapy. A quarter of a century later, that time has dropped to fifteen days. That is the finding of a 26-year cohort study published in The Lancet Regional Health – Europe, based on the French Hospital Database on HIV (ANRS CO4 FHDH), which includes 801 transgender women seen between 1997 and 2023.
801 trans women, 26 years of clinical records
The cohort includes trans women living with HIV who had never received antiretroviral treatment. Half were between 27 and 36 years old when they entered the study (median 31). The figure that frames the whole paper is place of birth: 65% (522 of 801) came from Latin America, and the rest from France or other countries. In other words, the clinical picture of this epidemic in France is largely that of migrant women.
From diagnosis to treatment: 3.9 months down to 15 days
The change in time to treatment is the central finding. In 1997-2004 the median was 3.9 months, with a very wide spread: one in four women waited more than two and a half years. In 2013-2023 the median fell to 0.5 months, that is, about fifteen days.
The full cascade moved in the same direction. Time to viral suppression after starting therapy fell from 6.2 months to 2.3 months. And the share of women on treatment with a suppressed viral load rose from 74.7% (109 of 146) to 95.5% (315 of 330) in the latest period, a level that meets the 95% target set by UNAIDS.
Why treatment started earlier also changed, and that shows how clinical practice has shifted. Until 2012 treatment was mostly brought forward for those who arrived in worse shape: having fewer than 200 CD4 cells or an AIDS diagnosis multiplied by more than three the probability of starting sooner. Between 2013 and 2023, however, the factor that speeds up initiation is a very high viral load (≥100,000 copies), a sign that care now follows clinical management criteria and guidelines rather than severity alone.
The gap that opens: country of birth
In the most recent period a new pattern appears. Being born in Latin America is associated with slower treatment initiation (hazard ratio 0.68; 95% confidence interval 0.50-0.91), and so is being born in other countries outside France (0.65; 0.45-0.95). In other words, among women with the same hospital care available, geographic origin predicts how long it takes to start.
The authors read this as a sign of structural barriers —administrative, social and economic— that clinical progress alone does not erase. Because most women in the cohort were born outside France, they conclude that targeted strategies and policies are needed to guarantee equitable and timely access to HIV care.
What it means for trans healthcare
- The French health system achieved real change: treatment within weeks and viral suppression in 95% of women on therapy.
- That success coexists with an inequality by origin: those born outside France take longer to start.
- The authors call for specific measures, because overall improvement does not distribute its benefits evenly.
Studies on transgender women living with HIV remain scarce in Europe, and this is one of the largest carried out so far: 26 years of follow-up and 801 clinical records. Its value lies in measuring precisely how much progress has been made and where the delay remains.
It is worth being precise about what the study examines and what it does not. It does not look at the origin of gender identity or the cause of gender dysphoria: it describes the healthcare received by transgender women and one inequality —country of birth— that runs through that care. For trans people and the teams treating them, the message is clear: the quality of a health system is also measured by who arrives late.
Source: The HIV care continuum and determinants of antiretroviral therapy initiation among transgender women in France (1997-2023): a longitudinal cohort study. Hemery J, Tattevin P, Marshall E, et al. The Lancet Regional Health – Europe 2026;69:101820. DOI: 10.1016/j.lanepe.2026.101820.
Article published at dosier.es. Image generated with artificial intelligence (marta ai → marta.syf.es).
